ME. W. CEOOKES ON EEPULSION EESULTING EEOM EADIATION. 
295 
comparison than if the same area of metal had been present in each vane, for in that 
case each alteration of curvature would have altered the distance between the glass 
and the edge of the vane. . 
315. These four radiometers were exhausted and tested with a standard candle 
3 inches from the centre of rotation. As the exhaustion approached the point of 
greatest sensitiveness, the fly A was the first to move, and it was in good rotation in 
the direction of the arrows before the others would rotate at all ; soon after A and B 
were rotating, C commenced to rotate, but it required the exhaustion to be carried to 
a higher point before I could get rotation in D. 
At the exhaustion corresponding to maximum sensitiveness, the rates of rotation 
were as follows : — - 
A made 2 1 ’4 revolutions a minute. 
B „ 15-0 
C „ 10-3 
D „ 4-6 
These figures show that the deeper the curvature, up to half a circle, the greater 
the repulsion. 
316. The action of dark heat was now tried, as in the experiments with the radio- 
meters, with “ favourably presented ” flat vanes. Dipped into hot water they behaved 
like the aluminium instruments (285, 286), rotating negatively whilst getting hot, 
then coming to rest, and rotating positively on cooling. 
317. Experiments were now tried with hot rings applied at the poles and at the 
equator of the bulbs, as in experiments 298, 299, 301. The results fully confirmed 
those then obtained, and afford an additional proof that the explanation given in 
pars. 302, 303 is correct. When the hot ring was applied above, negative rotation 
took place, and when the heat was applied only to the equator of the bulb, positive 
rotation took place. In all cases, on cooling, the movement was in the positive 
direction. 
318. The shape of the vanes was now again altered. It was found that cups were 
more easily affected by radiation than portions of cylinders, and that they were also more 
Fig. 23. 
easily fashioned into shape. Spherical moulds were prepared with radii of curvature 
respectively of 35, 14, and 10 millims. By means of these moulds, disks of gold foil, 
14‘5 millims. diameter, were bent in the form of shallow cups, brightly polished on 
both sides. They were mounted in pairs in the experimental bulb, fig. 4 (244), and 
